Big Tech · Satire
Meta Open Sources Model It Already Replaced with Better Model
Llama 3, released as open source six months after Llama 4 became Meta's production model, will be available for 'research and commercial use' by anyone who signs a license agreement.
MENLO PARK — Meta has announced the open source release of its Llama 3 language model, making it freely available to researchers and developers approximately six months after the company’s internal teams switched entirely to Llama 4.
“We believe in open AI for the benefit of humanity,” read an imagined company blog post, published underneath a stock photo of a very normal t-shirt.
Llama 3, which was Meta’s best model in Q2 of last year, is now approximately the fifteenth most capable publicly available model, sitting between two models from companies that no longer exist and one model that a graduate student fine-tuned on a laptop.
“This represents a significant contribution to the open source community,” said a Meta spokesperson, who confirmed that Llama 5 has been in production since January.
The release comes with a license that permits commercial use subject to 23 conditions, including a requirement that users not use the model to “demean or harm” individuals, which lawyers described as “technically enforceable but practically, you know, not really.”
Developers on social media greeted the release enthusiastically, noting that it was “better than nothing” and “probably fine for simple tasks.”
At press time, Meta had announced Llama 6.
